Presenter: James Benedict, Los Alamos National Laboratory
Title:
...
Improving the QBO through surrogate-accelerated parameter optimization and vertical grid modification
Abstract:
The purpose of MPAS-SI-MPM is to enhance the representation of sea-ice mechanics in E3SM using an elastic-decohesive constitutive model with the goal of ultimately improving coupled climate simulations and predictions. The more descriptive physics provides a mechanism to model leads explicitly, and to account for frazil ice and dense water formation in leads, for example. The improved representation will generate more realistic coupling fluxes to the atmosphere and ocean.
The model will be implemented using the material-point method (MPM). MPM provides a Lagrangian representation of sea ice which allows transport of state variables and tracers without numerical diffusion, as well as history along trajectories for constitutive models. GPU acceleration will be achieved in the sea-ice component of E3SM through modifications to the PUMIPic library from FASTMath.
Techniques for parameter calibration and model validation (especially for leads) will be developed. These techniques depend on new metrics for analyzing
spatiotemporal data with discontinuities, and/or lower dimensional featuresquasi-biennial oscillation (QBO) is the leading mode of variability in the tropical stratosphere. Though the QBO itself is confined to the equatorial region, it influences global climate patterns and weather extremes and regulates stratospheric chemical constituents. QBO representation in most global climate models remains poor. Particularly alarming are the large QBO biases in E3SM version 2 and development version 3. The E3SM biases are hypothesized to stem from an inadequate resolution of the vertical grid and poor representation of resolved and parameterized gravity waves, whose upward propagation from the troposphere into the stratosphere and associated momentum deposition drive the QBO. This presentation serves as a brief overview of our SciDAC project, which is focused on improving the QBO in E3SM and is comprised of the following ongoing or planned research tasks: (1) develop a machine learning-based workflow leveraging surrogate model accelerators to efficiently and objectively conduct parameter optimization in E3SM, (2) implement process-oriented diagnostics to advance our understanding of the QBO and evaluate whether the optimized QBO exhibits features consistent with physical processes inferred from observations, and (3) leverage E3SM configurations with alternative treatments of convection to more deeply probe the mechanisms that might be missing in the conventional E3SM. Preliminary project results that may be relevant to the E3SM v3 development cycle will also be discussed.
Date
Time
- PT: 8:30 am
- ET: 11:30 am
...
- web session: https://global.gotomeeting.com/join/570361173
- call number: (571) 317-3122 Access Code: 570-361-173, If busy, use alternate number: (773) 945-1029
Joining from a video-conferencing room or system? Dial: 67.217.95.2##570361173 , Cisco devices: 570361173@67.217.95.2
Attendees
- Mark Petersen
- Wuyin Lin
- Chris Golaz
- Jon Wolfe
- Ruby Leung
- Qi Tang
- Darin Comeau
- Michael J Prather
- Jim Benedict
...
| Page Properties | ||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
...